A cluster prediction model-based data collection for energy efficient wireless sensor network

Wireless sensor networks (WSN) are expected to cover the major portion of the earth's surface in the coming years. In the era of IoT, the WSN is the major data collection framework. To manage with the energy efficient data collection paradigm in WSN, numerous techniques have been suggested by the research community. In this paper, a data-aware energy conservation technique is proposed. Here, the inherent correlation between the consecutive observations of the sensor node and the data trend similarity between the neighboring sensor nodes are utilized to reduce the data transmission. A prediction-based data collection framework reduces the temporal data redundancy. ARIMA modeling is used to predict the data. The model is constructed by the (Clusterhead) CH node and is communicated to the cluster nodes. On every data collection round, the nodes compare the model predicted data and the observed data of the instant. If there is a deviation beyond the specified threshold, the nodes communicate the data difference to the CH. The data differences collected by the CH are compressed by using PCA technique. The compressed data are then sent to the sink node. Using this method, a huge portion of redundant data transmission is cut off. The method also maintains the collected data's accuracy within the predefined error threshold. Being a data reduction-based energy conservation technique, this results in reduced data collision. This method conserves 82% of energy with the error threshold of minimum level.